## ordered allocation sampler

Posted in Books, Statistics with tags , , , , , , , , , , , on November 29, 2021 by xi'an

Recently, Pierpaolo De Blasi and María Gil-Leyva arXived a proposal for a novel Gibbs sampler for mixture models. In both finite and infinite mixture models. In connection with Pitman (1996) theory of species sampling and with interesting features in terms of removing the vexing label switching features.

The key idea is to work with the mixture components in the random order of appearance in an exchangeable sequence from the mixing distribution (…) In accordance with the order of appearance, we derive a new Gibbs sampling algorithm that we name the ordered allocation sampler. “

This central idea is thus a reinterpretation of the mixture model as the marginal of the component model when its parameter is distributed as a species sampling variate. An ensuing marginal algorithm is to integrate out the weights and the allocation variables to only consider the non-empty component parameters and the partition function, which are label invariant. Which reminded me of the proposal we made in our 2000 JASA paper with Gilles Celeux and Merrilee Hurn (one of my favourite papers!). And of the [first paper in Statistical Methodology] 2004 partitioned importance sampling version with George Casella and Marty Wells. As in the later, the solution seems to require the prior on the component parameters to be conjugate (as I do not see a way to produce an unbiased estimator of the partition allocation probabilities).

The ordered allocation sample considers the posterior distribution of the different object made of the parameters and of the sequence of allocations to the components for the sample written in a given order, ie y¹,y², &tc. Hence y¹ always gets associated with component 1, y² with either component 1 or component 2, and so on. For this distribution, the full conditionals are available, incl. the full posterior on the number m of components, only depending on the data through the partition sizes and the number m⁺ of non-empty components. (Which relates to the debate as to whether or not m is estimable…) This sequential allocation reminded me as well of an earlier 2007 JRSS paper by Nicolas Chopin. Albeit using particles rather than Gibbs and applied to a hidden Markov model. Funny enough, their synthetic dataset univ4 almost resembles the Galaxy dataset (as in the above picture of mine)!

## latent variables for a hierarchical Poisson model

Posted in Books, Kids, pictures, Statistics, University life with tags , , , , , , , , on March 11, 2021 by xi'an

Answering a question on X validated about a rather standard hierarchical Poisson model, and its posterior Gibbs simulation, where observations are (d and w being a document and a word index, resp.)

$N_{w,d}\sim\mathcal P(\textstyle\sum_{1\le k\le K} \pi_{k,d}\varphi_{k,w})\qquad(1)$

I found myself dragged into an extended discussion on the validation of creating independent Poisson latent variables

$N_{k,w,d}\sim\mathcal P(\pi_{k,d}\varphi_{k,w})\qquad(2)$

since observing their sum in (1) was preventing the latent variables (2) from being independent. And then found out that the originator of the question had asked on X validated an unanswered and much more detailed question in 2016, even though the notations differ. The question does contain the solution I proposed above, including the Multinomial distribution on the Poisson latent variables given their sum (and the true parameters). As it should be since the derivation was done in a linked 2014 paper by Gopalan, Hofman, and Blei, later published in the Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence (UAI). I am thus bemused at the question resurfacing five years later in a much simplified version, but still exhibiting the same difficulty with the conditioning principles…

## averaged acceptance ratios

Posted in Statistics with tags , , , , , , , , , , , , , on January 15, 2021 by xi'an

In another recent arXival, Christophe Andrieu, Sinan Yıldırım, Arnaud Doucet, and Nicolas Chopin study the impact of averaging estimators of acceptance ratios in Metropolis-Hastings algorithms. (It is connected with the earlier arXival rephrasing Metropolis-Hastings in terms of involutions discussed here.)

“… it is possible to improve performance of this algorithm by using a modification where the acceptance ratio r(ξ) is integrated with respect to a subset of the proposed variables.”

This interpretation of the current proposal makes it a form of Rao-Blackwellisation, explicitly mentioned on p.18, where, using a mixture proposal, with an adapted acceptance probability, it depends on the integrated acceptance ratio only. Somewhat magically using this ratio and its inverse with probability ½. And it increases the average Metropolis-Hastings acceptance probability (albeit with a larger number of simulations). Since the ideal averaging is rarely available, the authors implement a Monte Carlo averaging version. With applications to the exchange algorithm and to reversible jump MCMC. The major application is to pseudo-marginal settings with a high complexity (in the number T of terms) and where the authors’ approach does scale efficiently with T. There is even an ABC side to the story as one illustration is made of the ABC approximation to the posterior of an α-stable sample. As an encompassing proposal for handling Metropolis-Hastings environments with latent variables and several versions of the acceptance ratios, this is quite an interesting paper that I think we will study in further detail with our students.

## mining gold [ABC in PNAS]

Posted in Books, Statistics with tags , , , , , , , , , , , on March 13, 2020 by xi'an

Johann Brehmer and co-authors have just published a paper in PNAS entitled “Mining gold from implicit models to improve likelihood-free inference”. (Besides the pun about mining gold, the paper also involves techniques named RASCAL and SCANDAL, respectively! For Ratio And SCore Approximate Likelihood ratio and SCore-Augmented Neural Density Approximates Likelihood.) This setup is not ABC per se in that their simulator is used both to generate training data and construct a tractable surrogate model. Exploiting Geyer’s (1994) classification trick of expressing the likelihood ratio as the optimal classification ratio when facing two equal-size samples from one density and the other.

“For all these inference strategies, the augmented data is particularly powerful for enhancing the power of simulation-based inference for small changes in the parameter θ.”

Brehmer et al. argue that “the most important novel contribution that differentiates our work from the existing methods is the observation that additional information can be extracted from the simulator, and the development of loss functions that allow us to use this “augmented” data to more efficiently learn surrogates for the likelihood function.” Rather than starting from a statistical model, they also seem to use a scientific simulator made of multiple layers of latent variables z, where

x=F⁰(u⁰,z¹,θ), z¹=G¹(u¹,z²), z²=G¹(u²,z³), …

although they also call the marginal of x, p(x|θ), an (intractable) likelihood.

“The integral of the log is not the log of the integral!”

The central notion behind the improvement is a form of Rao-Blackwellisation, exploiting the simulated z‘s. Joint score functions and joint likelihood ratios are then available. Ignoring biases, the authors demonstrate that the closest approximation to the joint likelihood ratio and the joint score function that only depends on x is the actual likelihood ratio and the actual score function, respectively. Which sounds like an older EM result, except that the roles of estimate and target quantity are somehow inverted: one is approximating the marginal with the joint, while the marginal is the “best” approximation of the joint. But in the implementation of the method, an estimate of the (observed and intractable) likelihood ratio is indeed produced towards minimising an empirical loss based on two simulated samples. Learning this estimate ê(x) then allows one to use it for the actual data. It however requires fitting a new ê(x) for each pair of parameters. Providing as well an estimator of the likelihood p(x|θ). (Hence the SCANDAL!!!) A second type of approximation of the likelihood starts from the approximate value of the likelihood p(x|θ⁰) at a fixed value θ⁰ and expands it locally as an exponential family shift, with the score t(x|θ⁰) as sufficient statistic.

I find the paper definitely interesting even though it requires the representation of the (true) likelihood as a marginalisation over multiple layers of latent variables z. And does not provide an evaluation of the error involved in the process when the model is misspecified. As a minor supplementary appeal of the paper, the use of an asymmetric Galton quincunx to illustrate an intractable array of latent variables will certainly induce me to exploit it in projects and courses!

[Disclaimer: I was not involved in the PNAS editorial process at any point!]

## unbiased product of expectations

Posted in Books, Statistics, University life with tags , , , , , , , , on August 5, 2019 by xi'an

While I was not involved in any way, or even aware of this research, Anthony Lee, Simone Tiberi, and Giacomo Zanella have an incoming paper in Biometrika, and which was partly written while all three authors were at the University of Warwick. The purpose is to design an efficient manner to approximate the product of n unidimensional expectations (or integrals) all computed against the same reference density. Which is not a real constraint. A neat remark that motivates the method in the paper is that an improved estimator can be connected with the permanent of the n x N matrix A made of the values of the n functions computed at N different simulations from the reference density. And involves N!/ (N-n)! terms rather than N to the power n. Since it is NP-hard to compute, a manageable alternative uses random draws from constrained permutations that are reasonably easy to simulate. Especially since, given that the estimator recycles most of the particles, it requires a much smaller version of N. Essentially N=O(n) with this scenario, instead of O(n²) with the basic Monte Carlo solution, towards a similar variance.

This framework offers many applications in latent variable models, including pseudo-marginal MCMC, of course, but also for ABC since the ABC posterior based on getting each simulated observation close enough from the corresponding actual observation fits this pattern (albeit the dependence on the chosen ordering of the data is an issue that can make the example somewhat artificial).